mothur.h

   1 #ifndef MOTHUR_H
   2 #define MOTHUR_H
   3
   4
   5
   6 /*
   7  *  mothur.h
   8  *  Mothur
   9  *
  10  *  Created by Sarah Westcott on 2/19/09.
  11  *  Copyright 2009 Schloss Lab UMASS Amherst. All rights reserved.
  12  *
  13  */
  14
  15 /* This file contains all the standard incudes we use in the project as well as some common utilities. */
  16
  17 //#include <cstddef>
  18
  19 //io libraries
  20 #include <iostream>
  21 #include <iomanip>
  22 #include <fstream>
  23 #include <sstream>
  24 #include <signal.h>
  25
  26
  27 //exception
  28 #include <stdexcept>
  29 #include <exception>
  30 #include <cstdlib>
  31
  32
  33 //containers
  34 #include <vector>
  35 #include <set>
  36 #include <map>
  37 #include <string>
  38 #include <list>
  39
  40 //math
  41 #include <cmath>
  42 #include <math.h>
  43 #include <algorithm>
  44
  45 //misc
  46 #include <cerrno>
  47 #include <ctime>
  48 #include <limits>
  49
  50 #ifdef USE_MPI
  51         #include "mpi.h"
  52 #endif
  53 /***********************************************************************/
  54
  55 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
  56         #include <sys/wait.h>
  57         #include <sys/time.h>
  58         #include <sys/resource.h>
  59         #include <unistd.h>
  60
  61         #ifdef USE_READLINE
  62                 #include <readline/readline.h>
  63                 #include <readline/history.h>
  64         #endif
  65
  66 #else
  67         #include <conio.h> //allows unbuffered screen capture from stdin
  68         #include <direct.h> //get cwd
  69         #include <windows.h>
  70         #include <psapi.h>
  71
  72 #endif
  73
  74 using namespace std;
  75
  76 #define exp(x) (exp((double) x))
  77 #define sqrt(x) (sqrt((double) x))
  78 #define log10(x) (log10((double) x))
  79 #define log2(x) (log10(x)/log10(2))
  80 #define isnan(x) ((x) != (x))
  81 #define isinf(x) (fabs(x) == std::numeric_limits<double>::infinity())
  82
  83 typedef unsigned long ull;
  84
  85 struct IntNode {
  86         int lvalue;
  87         int rvalue;
  88         int lcoef;
  89         int rcoef;
  90         IntNode* left;
  91         IntNode* right;
  92
  93         IntNode(int lv, int rv, IntNode* l, IntNode* r) : lvalue(lv), rvalue(rv), left(l), right(r) {};
  94         IntNode() {};
  95 };
  96
  97 struct ThreadNode {
  98         int* pid;
  99         IntNode* left;
 100         IntNode* right;
 101 };
 102
 103 /************************************************************/
 104 struct clusterNode {
 105         int numSeq;
 106         int parent;
 107         int smallChild; //used to make linkTable work with list and rabund. represents bin number of this cluster node
 108         clusterNode(int num, int par, int kid) : numSeq(num), parent(par), smallChild(kid) {};
 109 };
 110 /************************************************************/
 111 struct seqDist {
 112         int seq1;
 113         int seq2;
 114         float dist;
 115         seqDist() {}
 116         seqDist(int s1, int s2, float d) : seq1(s1), seq2(s2), dist(d) {}
 117         ~seqDist() {}
 118 };
 119 //********************************************************************************************************************
 120 //sorts lowest to highest
 121 inline bool compareSequenceDistance(seqDist left, seqDist right){
 122         return (left.dist < right.dist);
 123 }
 124 /***********************************************************************/
 125
 126 // snagged from http://www.parashift.com/c++-faq-lite/misc-technical-issues.html#faq-39.2
 127 // works for now, but there should be a way to do it without killing the whole program
 128
 129 class BadConversion : public runtime_error {
 130 public:
 131         BadConversion(const string& s) : runtime_error(s){ }
 132 };
 133
 134 //**********************************************************************************************************************
 135
 136 template<typename T>
 137 inline void convert(const string& s, T& x, bool failIfLeftoverChars = true){
 138         istringstream i(s);
 139         char c;
 140         if (!(i >> x) || (failIfLeftoverChars && i.get(c)))
 141                 throw BadConversion(s);
 142 }
 143
 144 //**********************************************************************************************************************
 145
 146 template<typename T>
 147 inline bool convertTestFloat(const string& s, T& x, bool failIfLeftoverChars = true){
 148         istringstream i(s);
 149         char c;
 150         if (!(i >> x) || (failIfLeftoverChars && i.get(c)))
 151         {
 152                 return false;
 153         }
 154         return true;
 155 }
 156
 157 //**********************************************************************************************************************
 158
 159 template<typename T>
 160 inline bool convertTest(const string& s, T& x, bool failIfLeftoverChars = true){
 161         istringstream i(s);
 162         char c;
 163         if (!(i >> x) || (failIfLeftoverChars && i.get(c)))
 164         {
 165                 cout << "unable to be converted into an integer.\n" << endl;
 166                 return false;
 167         }
 168         return true;
 169 }
 170
 171 //**********************************************************************************************************************
 172
 173 template<typename T>
 174 string toString(const T&x){
 175     stringstream output;
 176     output << x;
 177     return output.str();
 178 }
 179
 180 //**********************************************************************************************************************
 181
 182 template<typename T>
 183 string toHex(const T&x){
 184         stringstream output;
 185
 186         output << hex << x;
 187
 188     return output.str();
 189 }
 190 //**********************************************************************************************************************
 191
 192 template<typename T>
 193 string toString(const T&x, int i){
 194         stringstream output;
 195
 196         output.precision(i);
 197     output << fixed << x;
 198
 199     return output.str();
 200 }
 201 /***********************************************************************/
 202 inline int openOutputFileAppend(string fileName, ofstream& fileHandle){
 203
 204         fileHandle.open(fileName.c_str(), ios::app);
 205         if(!fileHandle) {
 206                 cout << "Error: Could not open " << fileName << endl;
 207                 return 1;
 208         }
 209         else {
 210                 return 0;
 211         }
 212
 213 }
 214 /***********************************************************************/
 215
 216 inline void gobble(istream& f){
 217
 218         char d;
 219     while(isspace(d=f.get()))           {;}
 220         f.putback(d);
 221
 222 }
 223 /***********************************************************************/
 224
 225 inline string getline(istringstream& fileHandle) {
 226         try {
 227
 228                 string line = "";
 229
 230                 while (!fileHandle.eof())       {
 231                         //get next character
 232                         char c = fileHandle.get();
 233
 234                         //are you at the end of the line
 235                         if ((c == '\n') || (c == '\r') || (c == '\f')){  break; }
 236                         else {          line += c;              }
 237                 }
 238
 239                 return line;
 240
 241         }
 242         catch(exception& e) {
 243                 cout << "Error in mothur function getline" << endl;
 244                 exit(1);
 245         }
 246 }
 247 /***********************************************************************/
 248
 249 inline string getline(ifstream& fileHandle) {
 250         try {
 251
 252                 string line = "";
 253
 254                 while (!fileHandle.eof())       {
 255                         //get next character
 256                         char c = fileHandle.get();
 257
 258                         //are you at the end of the line
 259                         if ((c == '\n') || (c == '\r') || (c == '\f')){  break; }
 260                         else {          line += c;              }
 261                 }
 262
 263                 return line;
 264
 265         }
 266         catch(exception& e) {
 267                 cout << "Error in mothur function getline" << endl;
 268                 exit(1);
 269         }
 270 }
 271 /***********************************************************************/
 272
 273 inline bool isTrue(string f){
 274
 275         if ((f == "TRUE") || (f == "T") || (f == "true") || (f == "t")) {       return true;    }
 276         else {  return false;  }
 277 }
 278
 279 /***********************************************************************/
 280
 281 inline float roundDist(float dist, int precision){
 282
 283         return int(dist * precision + 0.5)/float(precision);
 284
 285 }
 286 /***********************************************************************/
 287
 288 inline float ceilDist(float dist, int precision){
 289
 290         return int(ceil(dist * precision))/float(precision);
 291
 292 }
 293
 294 /***********************************************************************/
 295
 296 inline int getNumNames(string names){
 297
 298         int count = 0;
 299
 300         if(names != ""){
 301                 count = 1;
 302                 for(int i=0;i<names.size();i++){
 303                         if(names[i] == ','){
 304                                 count++;
 305                         }
 306                 }
 307         }
 308
 309         return count;
 310
 311 }
 312
 313 /**************************************************************************************************/
 314
 315 inline vector<vector<double> > binomial(int maxOrder){
 316
 317         vector<vector<double> > binomial(maxOrder+1);
 318
 319     for(int i=0;i<=maxOrder;i++){
 320                 binomial[i].resize(maxOrder+1);
 321                 binomial[i][0]=1;
 322                 binomial[0][i]=0;
 323     }
 324     binomial[0][0]=1;
 325
 326     binomial[1][0]=1;
 327     binomial[1][1]=1;
 328
 329     for(int i=2;i<=maxOrder;i++){
 330                 binomial[1][i]=0;
 331     }
 332
 333     for(int i=2;i<=maxOrder;i++){
 334                 for(int j=1;j<=maxOrder;j++){
 335                         if(i==j){       binomial[i][j]=1;                                                                       }
 336                         if(j>i) {       binomial[i][j]=0;                                                                       }
 337                         else    {       binomial[i][j]=binomial[i-1][j-1]+binomial[i-1][j];     }
 338                 }
 339     }
 340
 341         return binomial;
 342 }
 343
 344 /***********************************************************************/
 345
 346 inline string getRootName(string longName){
 347
 348         string rootName = longName;
 349
 350         if(longName.find_last_of(".") != longName.npos){
 351                 int pos = longName.find_last_of('.')+1;
 352                 rootName = longName.substr(0, pos);
 353         }
 354
 355         return rootName;
 356 }
 357 /***********************************************************************/
 358
 359 inline string getSimpleName(string longName){
 360
 361         string simpleName = longName;
 362
 363         size_t found;
 364         found=longName.find_last_of("/\\");
 365
 366         if(found != longName.npos){
 367                 simpleName = longName.substr(found+1);
 368         }
 369
 370                 //if(longName.find_last_of("/") != longName.npos){
 371                 //      int pos = longName.find_last_of('/')+1;
 372                 //      simpleName = longName.substr(pos, longName.length());
 373                 //}
 374
 375         return simpleName;
 376 }
 377
 378 /***********************************************************************/
 379
 380 inline int factorial(int num){
 381         int total = 1;
 382
 383         for (int i = 1; i <= num; i++) {
 384                 total *= i;
 385         }
 386
 387         return total;
 388 }
 389 /**************************************************************************************************
 390
 391 double min(double x, double y)
 392 {
 393     if(x<y){    return x;    }
 394     else   {    return y;    }
 395 }
 396
 397 /***********************************************************************/
 398
 399 inline string getPathName(string longName){
 400
 401         string rootPathName = longName;
 402
 403         if(longName.find_last_of("/\\") != longName.npos){
 404                 int pos = longName.find_last_of("/\\")+1;
 405                 rootPathName = longName.substr(0, pos);
 406         }
 407
 408         return rootPathName;
 409 }
 410 /***********************************************************************/
 411
 412 inline string hasPath(string longName){
 413
 414         string path = "";
 415
 416         size_t found;
 417         found=longName.find_last_of("~/\\");
 418
 419         if(found != longName.npos){
 420                 path = longName.substr(0, found+1);
 421         }
 422
 423         return path;
 424 }
 425
 426 /***********************************************************************/
 427
 428 inline string getExtension(string longName){
 429
 430         string extension = longName;
 431
 432         if(longName.find_last_of('.') != longName.npos){
 433                 int pos = longName.find_last_of('.');
 434                 extension = longName.substr(pos, longName.length());
 435         }
 436
 437         return extension;
 438 }
 439 /***********************************************************************/
 440 inline bool isBlank(string fileName){
 441
 442         ifstream fileHandle;
 443         fileHandle.open(fileName.c_str());
 444         if(!fileHandle) {
 445                 cout << "Error: Could not open " << fileName << endl;
 446                 return false;
 447         }else {
 448                 //check for blank file
 449                 gobble(fileHandle);
 450                 if (fileHandle.eof()) { fileHandle.close(); return true;  }
 451         }
 452         return false;
 453 }
 454 /***********************************************************************/
 455
 456 inline string getFullPathName(string fileName){
 457         try{
 458
 459         string path = hasPath(fileName);
 460         string newFileName;
 461         int pos;
 462
 463         if (path == "") { return fileName; } //its a simple name
 464         else { //we need to complete the pathname
 465                 // ex. ../../../filename
 466                 // cwd = /user/work/desktop
 467
 468                 string cwd;
 469                 //get current working directory
 470                 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
 471
 472                         if (path.find("~") != -1) { //go to home directory
 473                                 string homeDir = getenv ("HOME");
 474                                 newFileName = homeDir + fileName.substr(fileName.find("~")+1);
 475                                 return newFileName;
 476                         }else { //find path
 477                                 if (path.rfind("./") == -1) { return fileName; } //already complete name
 478                                 else { newFileName = fileName.substr(fileName.rfind("./")+2); } //save the complete part of the name
 479
 480                                 char* cwdpath = new char[1024];
 481
 482                                 size_t size;
 483                                 cwdpath=getcwd(cwdpath,size);
 484
 485                                 cwd = cwdpath;
 486
 487                                 //rip off first '/'
 488                                 string simpleCWD;
 489                                 if (cwd.length() > 0) { simpleCWD = cwd.substr(1); }
 490
 491                                 //break apart the current working directory
 492                                 vector<string> dirs;
 493                                 while (simpleCWD.find_first_of('/') != -1) {
 494                                         string dir = simpleCWD.substr(0,simpleCWD.find_first_of('/'));
 495                                         simpleCWD = simpleCWD.substr(simpleCWD.find_first_of('/')+1, simpleCWD.length());
 496                                         dirs.push_back(dir);
 497                                 }
 498                                 //get last one              // ex. ../../../filename = /user/work/desktop/filename
 499                                 dirs.push_back(simpleCWD);  //ex. dirs[0] = user, dirs[1] = work, dirs[2] = desktop
 500
 501
 502                                 int index = dirs.size()-1;
 503
 504                                 while((pos = path.rfind("./")) != -1) { //while you don't have a complete path
 505                                         if (pos == 0) { break;  //you are at the end
 506                                         }else if (path[(pos-1)] == '.') { //you want your parent directory ../
 507                                                 path = path.substr(0, pos-1);
 508                                                 index--;
 509                                                 if (index == 0) {  break; }
 510                                         }else if (path[(pos-1)] == '/') { //you want the current working dir ./
 511                                                 path = path.substr(0, pos);
 512                                         }else if (pos == 1) { break;  //you are at the end
 513                                         }else { cout << "cannot resolve path for " <<  fileName << endl; return fileName; }
 514                                 }
 515
 516                                 for (int i = index; i >= 0; i--) {
 517                                         newFileName = dirs[i] +  "/" + newFileName;
 518                                 }
 519
 520                                 newFileName =  "/" +  newFileName;
 521                                 return newFileName;
 522                         }
 523                 #else
 524                         if (path.find("~") != -1) { //go to home directory
 525                                 string homeDir = getenv ("HOMEPATH");
 526                                 newFileName = homeDir + fileName.substr(fileName.find("~")+1);
 527                                 return newFileName;
 528                         }else { //find path
 529                                 if (path.rfind(".\\") == -1) { return fileName; } //already complete name
 530                                 else { newFileName = fileName.substr(fileName.rfind(".\\")+2); } //save the complete part of the name
 531
 532                                 char *cwdpath = NULL;
 533                                 cwdpath = getcwd(NULL, 0); // or _getcwd
 534                                 if ( cwdpath != NULL) { cwd = cwdpath; }
 535                                 else { cwd = "";  }
 536
 537                                 //break apart the current working directory
 538                                 vector<string> dirs;
 539                                 while (cwd.find_first_of('\\') != -1) {
 540                                         string dir = cwd.substr(0,cwd.find_first_of('\\'));
 541                                         cwd = cwd.substr(cwd.find_first_of('\\')+1, cwd.length());
 542                                         dirs.push_back(dir);
 543
 544                                 }
 545                                 //get last one
 546                                 dirs.push_back(cwd);  //ex. dirs[0] = user, dirs[1] = work, dirs[2] = desktop
 547
 548                                 int index = dirs.size()-1;
 549
 550                                 while((pos = path.rfind(".\\")) != -1) { //while you don't have a complete path
 551                                         if (pos == 0) { break;  //you are at the end
 552                                         }else if (path[(pos-1)] == '.') { //you want your parent directory ../
 553                                                 path = path.substr(0, pos-1);
 554                                                 index--;
 555                                                 if (index == 0) {  break; }
 556                                         }else if (path[(pos-1)] == '\\') { //you want the current working dir ./
 557                                                 path = path.substr(0, pos);
 558                                         }else if (pos == 1) { break;  //you are at the end
 559                                         }else { cout << "cannot resolve path for " <<  fileName << endl; return fileName; }
 560                                 }
 561
 562                                 for (int i = index; i >= 0; i--) {
 563                                         newFileName = dirs[i] +  "\\" + newFileName;
 564                                 }
 565
 566                                 return newFileName;
 567                         }
 568
 569                 #endif
 570         }
 571         }
 572         catch(exception& e) {
 573                 cout << "Standard Error: " << e.what() << " has occurred in the mothur.h function getFullPathName. Please contact Pat Schloss at mothur.bugs@gmail.com." << "\n";
 574                 exit(1);
 575         }
 576 }
 577 /***********************************************************************/
 578
 579 inline int openInputFile(string fileName, ifstream& fileHandle, string m){
 580
 581         //get full path name
 582         string completeFileName = getFullPathName(fileName);
 583
 584         fileHandle.open(completeFileName.c_str());
 585         if(!fileHandle) {
 586                 return 1;
 587         }else {
 588                 //check for blank file
 589                 gobble(fileHandle);
 590                 return 0;
 591         }
 592 }
 593 /***********************************************************************/
 594
 595 inline int openInputFile(string fileName, ifstream& fileHandle){
 596
 597         //get full path name
 598         string completeFileName = getFullPathName(fileName);
 599
 600         fileHandle.open(completeFileName.c_str());
 601         if(!fileHandle) {
 602                 cout << "Error: Could not open " << completeFileName << endl;
 603                 return 1;
 604         }
 605         else {
 606                 //check for blank file
 607                 gobble(fileHandle);
 608                 if (fileHandle.eof()) { cout << completeFileName << " is blank. Please correct." << endl;   }
 609
 610                 return 0;
 611         }
 612
 613 }
 614 /***********************************************************************/
 615
 616 inline int renameFile(string oldName, string newName){
 617
 618         ifstream inTest;
 619         int exist = openInputFile(newName, inTest, "");
 620
 621 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
 622         if (exist == 0) { //you could open it so you want to delete it
 623                 inTest.close();
 624                 string command = "rm " + newName;
 625                 system(command.c_str());
 626         }
 627
 628         string command = "mv " + oldName + " " + newName;
 629         system(command.c_str());
 630 #else
 631         remove(newName.c_str());
 632         int renameOk = rename(oldName.c_str(), newName.c_str());
 633 #endif
 634         return 0;
 635 }
 636
 637 /***********************************************************************/
 638
 639 inline int openOutputFile(string fileName, ofstream& fileHandle){
 640
 641         string completeFileName = getFullPathName(fileName);
 642
 643         fileHandle.open(completeFileName.c_str(), ios::trunc);
 644         if(!fileHandle) {
 645                 cout << "Error: Could not open " << completeFileName << endl;
 646                 return 1;
 647         }
 648         else {
 649                 return 0;
 650         }
 651
 652 }
 653
 654 /***********************************************************************/
 655
 656 inline int getNumSeqs(ifstream& file){
 657
 658         int numSeqs = count(istreambuf_iterator<char>(file),istreambuf_iterator<char>(), '>');
 659         file.seekg(0);
 660         return numSeqs;
 661
 662 }
 663 /***********************************************************************/
 664 inline void getNumSeqs(ifstream& file, int& numSeqs){
 665
 666         string input;
 667         numSeqs = 0;
 668         while(!file.eof()){
 669                 input = getline(file);
 670                 if (input.length() != 0) {
 671                         if(input[0] == '>'){ numSeqs++; }
 672                 }
 673         }
 674 }
 675
 676 /***********************************************************************/
 677
 678 inline bool inVector(string member, vector<string> group){
 679
 680         for (int i = 0; i < group.size(); i++) {
 681                 if (group[i] == member) {  return true;         }
 682         }
 683
 684         return false;
 685 }
 686 /***********************************************************************/
 687
 688 //This function parses the estimator options and puts them in a vector
 689 inline void splitAtChar(string& estim, vector<string>& container, char symbol) {
 690         try {
 691                 string individual;
 692
 693                 while (estim.find_first_of(symbol) != -1) {
 694                         individual = estim.substr(0,estim.find_first_of(symbol));
 695                         if ((estim.find_first_of(symbol)+1) <= estim.length()) { //checks to make sure you don't have dash at end of string
 696                                 estim = estim.substr(estim.find_first_of(symbol)+1, estim.length());
 697                                 container.push_back(individual);
 698                         }
 699                 }
 700                 //get last one
 701                 container.push_back(estim);
 702         }
 703         catch(exception& e) {
 704                 cout << "Standard Error: " << e.what() << " has occurred in the mothur.h function splitAtDash. Please contact Pat Schloss at mothur.bugs@gmail.com." << "\n";
 705                 exit(1);
 706         }
 707 }
 708
 709 /***********************************************************************/
 710
 711 //This function parses the estimator options and puts them in a vector
 712 inline void splitAtDash(string& estim, vector<string>& container) {
 713         try {
 714                 string individual;
 715
 716                 while (estim.find_first_of('-') != -1) {
 717                         individual = estim.substr(0,estim.find_first_of('-'));
 718                         if ((estim.find_first_of('-')+1) <= estim.length()) { //checks to make sure you don't have dash at end of string
 719                                 estim = estim.substr(estim.find_first_of('-')+1, estim.length());
 720                                 container.push_back(individual);
 721                         }
 722                 }
 723                 //get last one
 724                 container.push_back(estim);
 725         }
 726         catch(exception& e) {
 727                 cout << "Standard Error: " << e.what() << " has occurred in the mothur.h function splitAtDash. Please contact Pat Schloss at mothur.bugs@gmail.com." << "\n";
 728                 exit(1);
 729         }
 730 }
 731
 732 /***********************************************************************/
 733 //This function parses the label options and puts them in a set
 734 inline void splitAtDash(string& estim, set<string>& container) {
 735         try {
 736                 string individual;
 737
 738                 while (estim.find_first_of('-') != -1) {
 739                         individual = estim.substr(0,estim.find_first_of('-'));
 740                         if ((estim.find_first_of('-')+1) <= estim.length()) { //checks to make sure you don't have dash at end of string
 741                                 estim = estim.substr(estim.find_first_of('-')+1, estim.length());
 742                                 container.insert(individual);
 743                         }
 744                 }
 745                 //get last one
 746                 container.insert(estim);
 747         }
 748         catch(exception& e) {
 749                 cout << "Standard Error: " << e.what() << " has occurred in the mothur.h function splitAtDash. Please contact Pat Schloss at mothur.bugs@gmail.com." << "\n";
 750                 exit(1);
 751         }
 752 }
 753 /***********************************************************************/
 754 //This function parses the line options and puts them in a set
 755 inline void splitAtDash(string& estim, set<int>& container) {
 756         try {
 757                 string individual;
 758                 int lineNum;
 759
 760                 while (estim.find_first_of('-') != -1) {
 761                         individual = estim.substr(0,estim.find_first_of('-'));
 762                         if ((estim.find_first_of('-')+1) <= estim.length()) { //checks to make sure you don't have dash at end of string
 763                                 estim = estim.substr(estim.find_first_of('-')+1, estim.length());
 764                                 convert(individual, lineNum); //convert the string to int
 765                                 container.insert(lineNum);
 766                         }
 767                 }
 768                 //get last one
 769                 convert(estim, lineNum); //convert the string to int
 770                 container.insert(lineNum);
 771         }
 772         catch(exception& e) {
 773                 cout << "Standard Error: " << e.what() << " has occurred in the mothur.h function splitAtDash. Please contact Pat Schloss at mothur.bugs@gmail.com." << "\n";
 774                 exit(1);
 775         }
 776 }
 777 /***********************************************************************/
 778 //This function parses the a string and puts peices in a vector
 779 inline void splitAtComma(string& estim, vector<string>& container) {
 780         try {
 781                 string individual;
 782
 783                 while (estim.find_first_of(',') != -1) {
 784                         individual = estim.substr(0,estim.find_first_of(','));
 785                         if ((estim.find_first_of(',')+1) <= estim.length()) { //checks to make sure you don't have comma at end of string
 786                                 estim = estim.substr(estim.find_first_of(',')+1, estim.length());
 787                                 container.push_back(individual);
 788                         }
 789                 }
 790                 //get last one
 791                 container.push_back(estim);
 792         }
 793         catch(exception& e) {
 794                 cout << "Standard Error: " << e.what() << " has occurred in the mothur.h function splitAtComma. Please contact Pat Schloss at mothur.bugs@gmail.com." << "\n";
 795                 exit(1);
 796         }
 797 }
 798 /***********************************************************************/
 799
 800 //This function splits up the various option parameters
 801 inline void splitAtComma(string& prefix, string& suffix){
 802         try {
 803                 prefix = suffix.substr(0,suffix.find_first_of(','));
 804                 if ((suffix.find_first_of(',')+2) <= suffix.length()) {  //checks to make sure you don't have comma at end of string
 805                         suffix = suffix.substr(suffix.find_first_of(',')+1, suffix.length());
 806                         string space = " ";
 807                         while(suffix.at(0) == ' ')
 808                                 suffix = suffix.substr(1, suffix.length());
 809                 }
 810
 811         }
 812         catch(exception& e) {
 813                 cout << "Standard Error: " << e.what() << " has occurred in the mothur.h function splitAtComma. Please contact Pat Schloss at mothur.bugs@gmail.com." << "\n";
 814                 exit(1);
 815         }
 816 }
 817 /***********************************************************************/
 818
 819 //This function separates the key value from the option value i.e. dist=96_...
 820 inline void splitAtEquals(string& key, string& value){
 821         try {
 822                 if(value.find_first_of('=') != -1){
 823                         key = value.substr(0,value.find_first_of('='));
 824                         if ((value.find_first_of('=')+1) <= value.length()) {
 825                                 value = value.substr(value.find_first_of('=')+1, value.length());
 826                         }
 827                 }else{
 828                         key = value;
 829                         value = 1;
 830                 }
 831         }
 832         catch(exception& e) {
 833                 cout << "Standard Error: " << e.what() << " has occurred in the mothur.h function splitAtEquals. Please contact Pat Schloss at mothur.bugs@gmail.com." << "\n";
 834                 exit(1);
 835         }
 836 }
 837
 838 /**************************************************************************************************/
 839
 840 inline bool inUsersGroups(string groupname, vector<string> Groups) {
 841         try {
 842                 for (int i = 0; i < Groups.size(); i++) {
 843                         if (groupname == Groups[i]) { return true; }
 844                 }
 845                 return false;
 846         }
 847         catch(exception& e) {
 848                 cout << "Standard Error: " << e.what() << " has occurred in the mothur.h function inUsersGroups. Please contact Pat Schloss at mothur.bugs@gmail.com." << "\n";
 849                 exit(1);
 850         }
 851 }
 852 /**************************************************************************************************/
 853 //returns true if any of the strings in first vector are in second vector
 854 inline bool inUsersGroups(vector<string> groupnames, vector<string> Groups) {
 855         try {
 856
 857                 for (int i = 0; i < groupnames.size(); i++) {
 858                         if (inUsersGroups(groupnames[i], Groups)) { return true; }
 859                 }
 860                 return false;
 861         }
 862         catch(exception& e) {
 863                 cout << "Standard Error: " << e.what() << " has occurred in the mothur.h function inUsersGroups. Please contact Pat Schloss at mothur.bugs@gmail.com." << "\n";
 864                 exit(1);
 865         }
 866 }
 867 /***********************************************************************/
 868 //this function determines if the user has given us labels that are smaller than the given label.
 869 //if so then it returns true so that the calling function can run the previous valid distance.
 870 //it's a "smart" distance function.  It also checks for invalid labels.
 871 inline bool anyLabelsToProcess(string label, set<string>& userLabels, string errorOff) {
 872         try {
 873
 874                 set<string>::iterator it;
 875                 vector<float> orderFloat;
 876                 map<string, float> userMap;  //the conversion process removes trailing 0's which we need to put back
 877                 map<string, float>::iterator it2;
 878                 float labelFloat;
 879                 bool smaller = false;
 880
 881                 //unique is the smallest line
 882                 if (label == "unique") {  return false;  }
 883                 else {
 884                         if (convertTestFloat(label, labelFloat)) {
 885                                 convert(label, labelFloat);
 886                         }else { //cant convert
 887                                 return false;
 888                         }
 889                 }
 890
 891                 //go through users set and make them floats
 892                 for(it = userLabels.begin(); it != userLabels.end(); ++it) {
 893
 894                         float temp;
 895                         if ((*it != "unique") && (convertTestFloat(*it, temp) == true)){
 896                                 convert(*it, temp);
 897                                 orderFloat.push_back(temp);
 898                                 userMap[*it] = temp;
 899                         }else if (*it == "unique") {
 900                                 orderFloat.push_back(-1.0);
 901                                 userMap["unique"] = -1.0;
 902                         }else {
 903                                 if (errorOff == "") {  cout << *it << " is not a valid label." << endl;  }
 904                                 userLabels.erase(*it);
 905                                 it--;
 906                         }
 907                 }
 908
 909                 //sort order
 910                 sort(orderFloat.begin(), orderFloat.end());
 911
 912                 /*************************************************/
 913                 //is this label bigger than any of the users labels
 914                 /*************************************************/
 915
 916                 //loop through order until you find a label greater than label
 917                 for (int i = 0; i < orderFloat.size(); i++) {
 918                         if (orderFloat[i] < labelFloat) {
 919                                 smaller = true;
 920                                 if (orderFloat[i] == -1) {
 921                                         if (errorOff == "") { cout << "Your file does not include the label unique." << endl; }
 922                                         userLabels.erase("unique");
 923                                 }
 924                                 else {
 925                                         if (errorOff == "") { cout << "Your file does not include the label " << endl; }
 926                                         string s = "";
 927                                         for (it2 = userMap.begin(); it2!= userMap.end(); it2++) {
 928                                                 if (it2->second == orderFloat[i]) {
 929                                                         s = it2->first;
 930                                                         //remove small labels
 931                                                         userLabels.erase(s);
 932                                                         break;
 933                                                 }
 934                                         }
 935                                         if (errorOff == "") {cout << s <<  ". I will use the next smallest distance. " << endl; }
 936                                 }
 937                         //since they are sorted once you find a bigger one stop looking
 938                         }else { break; }
 939                 }
 940
 941                 return smaller;
 942
 943         }
 944         catch(exception& e) {
 945                 cout << "Standard Error: " << e.what() << " has occurred in the mothur.h function anyLabelsToProcess. Please contact Pat Schloss at mothur.bugs@gmail.com." << "\n";
 946                 exit(1);
 947         }
 948 }
 949
 950 /**************************************************************************************************/
 951 inline void appendFiles(string temp, string filename) {
 952         try{
 953                 ofstream output;
 954                 ifstream input;
 955
 956                 //open output file in append mode
 957                 openOutputFileAppend(filename, output);
 958                 int ableToOpen = openInputFile(temp, input, "no error");
 959
 960                 if (ableToOpen == 0) { //you opened it
 961                         while(char c = input.get()){
 962                                 if(input.eof())         {       break;                  }
 963                                 else                            {       output << c;    }
 964                         }
 965                         input.close();
 966                 }
 967
 968                 output.close();
 969         }
 970         catch(exception& e) {
 971                 cout << "Standard Error: " << e.what() << " has occurred in the mothur.h function appendFiles. Please contact Pat Schloss at mothur.bugs@gmail.com." << "\n";
 972                 exit(1);
 973         }
 974 }
 975
 976 /**************************************************************************************************/
 977 inline string sortFile(string distFile, string outputDir){
 978         try {
 979
 980                 //if (outputDir == "") {  outputDir += hasPath(distFile);  }
 981                 string outfile = getRootName(distFile) + "sorted.dist";
 982
 983
 984                 //if you can, use the unix sort since its been optimized for years
 985                 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
 986                         string command = "sort -n -k +3 " + distFile + " -o " + outfile;
 987                         system(command.c_str());
 988                 #else //you are stuck with my best attempt...
 989                         //windows sort does not have a way to specify a column, only a character in the line
 990                         //since we cannot assume that the distance will always be at the the same character location on each line
 991                         //due to variable sequence name lengths, I chose to force the distance into first position, then sort and then put it back.
 992
 993                         //read in file line by file and put distance first
 994                         string tempDistFile = distFile + ".temp";
 995                         ifstream input;
 996                         ofstream output;
 997                         openInputFile(distFile, input);
 998                         openOutputFile(tempDistFile, output);
 999
1000                         string firstName, secondName;
1001                         float dist;
1002                         while (input) {
1003                                 input >> firstName >> secondName >> dist;
1004                                 output << dist << '\t' << firstName << '\t' << secondName << endl;
1005                                 gobble(input);
1006                         }
1007                         input.close();
1008                         output.close();
1009
1010
1011                         //sort using windows sort
1012                         string tempOutfile = outfile + ".temp";
1013                         string command = "sort " + tempDistFile + " /O " + tempOutfile;
1014                         system(command.c_str());
1015
1016                         //read in sorted file and put distance at end again
1017                         ifstream input2;
1018                         openInputFile(tempOutfile, input2);
1019                         openOutputFile(outfile, output);
1020
1021                         while (input2) {
1022                                 input2 >> dist >> firstName >> secondName;
1023                                 output << firstName << '\t' << secondName << '\t' << dist << endl;
1024                                 gobble(input2);
1025                         }
1026                         input2.close();
1027                         output.close();
1028
1029                         //remove temp files
1030                         remove(tempDistFile.c_str());
1031                         remove(tempOutfile.c_str());
1032                 #endif
1033
1034                 return outfile;
1035         }
1036         catch(exception& e) {
1037                 cout << "Standard Error: " << e.what() << " has occurred in the mothur.h function sortfile. Please contact Pat Schloss at mothur.bugs@gmail.com." << "\n";
1038                 exit(1);
1039         }
1040 }
1041 /**************************************************************************************************/
1042 inline vector<unsigned long int> setFilePosFasta(string filename, int& num) {
1043
1044                         vector<unsigned long int> positions;
1045                         ifstream inFASTA;
1046                         openInputFile(filename, inFASTA);
1047
1048                         string input;
1049                         while(!inFASTA.eof()){
1050                                 input = getline(inFASTA); gobble(inFASTA);
1051                                 if (input.length() != 0) {
1052                                         if(input[0] == '>'){    unsigned long int pos = inFASTA.tellg(); positions.push_back(pos - input.length() - 1); }
1053                                 }
1054                         }
1055                         inFASTA.close();
1056
1057                         num = positions.size();
1058
1059                         /*FILE * pFile;
1060                         long size;
1061
1062                         //get num bytes in file
1063                         pFile = fopen (filename.c_str(),"rb");
1064                         if (pFile==NULL) perror ("Error opening file");
1065                         else{
1066                                 fseek (pFile, 0, SEEK_END);
1067                                 size=ftell (pFile);
1068                                 fclose (pFile);
1069                         }*/
1070
1071                         unsigned long int size = positions[(positions.size()-1)];
1072                         ifstream in;
1073                         openInputFile(filename, in);
1074
1075                         in.seekg(size);
1076
1077                         while(char c = in.get()){
1078                                 if(in.eof())            {       break;  }
1079                                 else                            {       size++; }
1080                         }
1081                         in.close();
1082
1083                         positions.push_back(size);
1084
1085                         return positions;
1086 }
1087 /**************************************************************************************************/
1088 inline vector<unsigned long int> setFilePosEachLine(string filename, int& num) {
1089
1090                         vector<unsigned long int> positions;
1091                         ifstream in;
1092                         openInputFile(filename, in);
1093
1094                         string input;
1095                         while(!in.eof()){
1096                                 unsigned long int lastpos = in.tellg();
1097                                 input = getline(in); gobble(in);
1098                                 if (input.length() != 0) {
1099                                         unsigned long int pos = in.tellg();
1100                                         if (pos != -1) { positions.push_back(pos - input.length() - 1); }
1101                                         else {  positions.push_back(lastpos);  }
1102                                 }
1103                         }
1104                         in.close();
1105
1106                         num = positions.size();
1107
1108                         FILE * pFile;
1109                         unsigned long int size;
1110
1111                         //get num bytes in file
1112                         pFile = fopen (filename.c_str(),"rb");
1113                         if (pFile==NULL) perror ("Error opening file");
1114                         else{
1115                                 fseek (pFile, 0, SEEK_END);
1116                                 size=ftell (pFile);
1117                                 fclose (pFile);
1118                         }
1119
1120                         positions.push_back(size);
1121
1122                         return positions;
1123 }
1124 /**************************************************************************************************/
1125 inline bool checkReleaseVersion(ifstream& file, string version) {
1126         try {
1127
1128                 bool good = true;
1129
1130                 string line = getline(file);
1131
1132                 //before we added this check
1133                 if (line[0] != '#') {  good = false;  }
1134                 else {
1135                         //rip off #
1136                         line = line.substr(1);
1137
1138                         vector<string> versionVector;
1139                         splitAtChar(version, versionVector, '.');
1140
1141                         //check file version
1142                         vector<string> linesVector;
1143                         splitAtChar(line, linesVector, '.');
1144
1145                         if (versionVector.size() != linesVector.size()) { good = false; }
1146                         else {
1147                                 for (int j = 0; j < versionVector.size(); j++) {
1148                                         int num1, num2;
1149                                         convert(versionVector[j], num1);
1150                                         convert(linesVector[j], num2);
1151
1152                                         //if mothurs version is newer than this files version, then we want to remake it
1153                                         if (num1 > num2) {  good = false; break;  }
1154                                 }
1155                         }
1156
1157                 }
1158
1159                 if (!good) {  file.close();  }
1160                 else { file.seekg(0);  }
1161
1162                 return good;
1163         }
1164         catch(exception& e) {
1165                 cout << "Standard Error: " << e.what() << " has occurred in the mothur.h function checkReleaseVersion. Please contact Pat Schloss at mothur.bugs@gmail.com." << "\n";
1166                 exit(1);
1167         }
1168 }
1169 /**************************************************************************************************/
1170 #endif
1171